Comparison of simulation-based algorithms for parameter estimation and state reconstruction in nonlinear state-space models

This study aims at comparing simulation-based approaches for estimating both the state and unknown parameters in nonlinear state-space models. Numerical results on different toy models show that combination of a Conditional Particle Filter (CPF) with Backward Simulation (BS) smoother Stochastic Expectation-Maximization (SEM) algorithm is promising approach. The CPFBS run small number particles allows to explore efficiently simulate relevant trajectories conditionally observations. When combined SEM algorithm, this provides accurate estimates models, where application EM algorithms standard particle or an ensemble Kalman limited.